๐Ÿ“ฆ EqualifyEverything / equalify-reflow

๐Ÿ“„ prompt_improvement.yml ยท 58 lines
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58name: Prompt / agent improvement
description: Improve accuracy or behaviour of an AI agent in the pipeline
labels: ["prompt", "agents"]
body:
  - type: markdown
    attributes:
      value: |
        Equalify Reflow's pipeline uses eight PydanticAI agents for structure analysis, heading reconciliation, page content correction, boundary repair, code tagging, footnote relocation, and more. See [AGENTS.md](../../AGENTS.md) for the full list and how to iterate on them.

        Use this template if you have a concrete example where a pipeline step produced an incorrect or suboptimal output and you want to propose a prompt change.
  - type: dropdown
    id: agent
    attributes:
      label: Which agent / pipeline phase?
      options:
        - Structure analysis (Phase 2)
        - Heading reconciliation (Phase 3)
        - Heading levels (Phase 4)
        - Page content corrections (Phase 5) โ€” main
        - Page content corrections (Phase 5) โ€” table subagent
        - Page content corrections (Phase 5) โ€” list subagent
        - Page content corrections (Phase 5) โ€” image subagent
        - Code block languages (Phase 6)
        - Cross-page boundary fixes (Phase 7) โ€” main
        - Cross-page boundary fixes (Phase 7) โ€” footnote relocation
        - Other / unsure
    validations:
      required: true
  - type: textarea
    id: input
    attributes:
      label: Input (document or excerpt)
      description: What was the input? A public-domain PDF or an anonymised excerpt is ideal.
    validations:
      required: true
  - type: textarea
    id: actual
    attributes:
      label: Actual output
      description: What did the pipeline produce?
      render: markdown
    validations:
      required: true
  - type: textarea
    id: expected
    attributes:
      label: Expected output
      description: What should it have produced, and why?
      render: markdown
    validations:
      required: true
  - type: textarea
    id: prompt_change
    attributes:
      label: Proposed prompt change
      description: Optional. If you have already drafted a prompt change, share it here. Otherwise leave blank and the maintainers will iterate.
      render: text